hi there, I have a strange cluster split:
* 3 Nodes using 2.10.5 and Appia
* Backend on Node 1 was disabled cause of inconsistency
2007-08-29 09:18:14,906 WARN controller.RequestManager.botdb Disabling all
backends after an inconsistency was detected for request 281474976717685
281474976710662 delete from ...
2007-08-29 09:18:14,952 INFO sequoia.controller.connection 1 connections freed
on "jdbc:postgresql://localhost003048888BCE:54310/botdb"
2007-08-29 09:18:14,953 WARN sequoia.controller.connection Some connections
are still active, destroying them.
2007-08-29 09:18:15,058 WARN controller.loadbalancer.RAIDb1 Active
transactions after backend botdb1_003048888BCE is disabled: [281474976710662]
2007-08-29 09:18:15,059 INFO controller.RequestManager.botdb Backend
botdb1_003048888BCE is now disabled
* Cluster works for the following days, generating error logs, cause of
disabled backend on one node, but there seems to be no problem
* Backup works every night and even on disabled node the dump from
another node is fetched without error
* After some days, this dump fetch doesn't work on the disabled node and
the cluster is split after this.
2007-09-02 00:26:27,271 WARN controller.RequestManager.botdb SQLException
while executing distributed abort
java.sql.SQLException: Transaction 1688849860263940 is not active, rejecting
the rollback.
at
org.continuent.sequoia.controller.scheduler.AbstractScheduler.rollback(AbstractScheduler.java:1244)
at
org.continuent.sequoia.controller.virtualdatabase.protocol.DistributedAbort.executeCommand(DistributedAbort.java:127)
at
org.continuent.sequoia.controller.virtualdatabase.protocol.DistributedTransactionMarker.handleMessageMultiThreaded(DistributedTransactionMarker.java:131)
at
org.continuent.sequoia.controller.virtualdatabase.DistributedVirtualDatabase.handleMessageMultiThreaded(DistributedVirtualDatabase.java:357)
at
org.continuent.hedera.adapters.MulticastRequestAdapterThread.run(MulticastRequestAdapterThread.java:102)
2007-09-02 00:26:27,282 INFO controller.virtualdatabase.botdb Checkpoint
backup dbdump-2007-09-02-00-25-21-192.168.192.102:25322-20070902002522242+0200
was stored
2007-09-02 00:26:27,282 INFO controller.virtualdatabase.botdb Backend
botdb1_003048886032 disabled on controller Member(address=/193.8.106.102:21080,
uid=193.8.106.102:21080)
2007-09-02 00:27:04,985 ERROR sequoia.controller.recoverylog Recovery log was
unable to update request completion status: [EMAIL PROTECTED] RECOVERY SET
exec_status=?,update_count=?,exec_time=? WHERE log_id=?], parameters=[[S],
[-1], [0], [1504]]]
2007-09-02 00:31:57,834 WARN sequoia.controller.scheduler Waiting for 1
pending writes
2007-09-02 00:32:27,844 WARN sequoia.controller.scheduler Waiting for 1
pending writes
2007-09-02 00:33:27,864 WARN sequoia.controller.scheduler Waiting for 1
pending writes
2007-09-02 00:35:27,891 WARN sequoia.controller.scheduler Waiting for 1
pending writes
2007-09-02 00:35:54,493 INFO continuent.hedera.gms
Member(address=/193.8.106.102:21080, uid=193.8.106.102:21080) failed in
Group(gid=botdb)
2007-09-02 00:35:54,526 WARN controller.virtualdatabase.botdb Controller
Member(address=/193.8.106.102:21080, uid=193.8.106.102:21080) has left the
cluster.
Some Questions:
* Why is the dump fetched to the node with the disabled backend and why
does it work? The backend is disabled, so fetching the dump and
synchronizing recovery log should not be done
* Any idea why it failed this time and the whole cluster was split off
after it?
Thanx in advance
Stefan
--
Zertificon Solutions GmbH
Landsberger Allee 117, 10407 Berlin, Germany
GF: Herbert Nebel, Dr. Burkhard Wiegel
HRB 94059, AG Berlin-Charlottenburg
http://www.zertificon.com
https://www.globaltrustpoint.com/[EMAIL PROTECTED]
+49 (0)30-5900 300-0 (fax -99)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Z1 SecureMail" by Zertificon
...the leading server solutions for Secure & Trustable E-Mail
Try our Policy controlled S/MIME & OpenPGP & HTTPS Messaging!!
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ Sequoia mailing list [email protected] https://forge.continuent.org/mailman/listinfo/sequoia
